DIFFRACTION-LIMITED POLARIMETRIC IMAGING OF PROTOPLANETARY DISKS AND MASS-LOSS SHELLS

WITH VAMPIRES Sydney Institute for Astronomy, University of Sydney

Subaru Telescope, National Astronomical Observatory of Japan The Australian Astronomical Observatory (AAO)

MQ Photonics Research Centre, Macquarie University

Image: NASA/JPL-Caltech

BACKGROUND - THE VAMPIRES INSTRUMENT▸ Polarimetric aperture masking and imaging in the visible

▸ Aperture masking- Diffraction limited imaging through the atmosphere… …resolution of 10s mas

▸ Differential polarimetry - imaging scattered starlight

▸ Precision differential calibration beats the contrast problem (triple-differential)

▸ Measurement of dust grain type, size

▸ ‘Hitch-hiker instrument’ - operates on visible channel of SCExAO on Subaru telescope, simultaneous with IR observations (e.g. CHARIS)

SCIENCE CASE: IMAGING THE INNER REGION OF PROTO-PLANETARY DISKS AT SOLAR-SYSTEM SCALES

Inner region (10’s of AU) critical for understanding planet formation and disk structure • Precise measurement of transition disk gaps and

inner disk • Density perturbations from massive bodies within

disk (e.g. warps, spirals) • Gravitationally bound clumping dust immediately

surrounding forming planets • Dust grain species and size determination

Complementary to coronagraphy • Resolution ~10 mas; 0.5λ/D

• ~1 AU at 100 pc • NRM F.O.V. ~500 mas, PDI ~3 arcsec • Scattered starlight, not thermal

MWC758 - Base image: Grady et al., ApJ, 2013

BackgroundCircumstellar disks naturally arise in simulations of the gravitational collapse of large clouds due to the conserva-

tion of angular momentum. There are many lines of independent evidence (such as the regularity of our solar systemnoted by Kant) which point to such disks being the cradle for planetary birth. In recent years a picture for some, butby no means all, of the steps needed for the formation of a mature solar system has been emerging.

Within a typical timescale of 0.5 Myr6 a visible star emerges from the dusty cocoon of its birth (evolving from aClass I to Class II young stellar object), and accretion onto the central star is effectively complete leaving a remnantdisk with only about⇠5 MJupiter (or about 1 % of the mass of the central star)7. The dust, which dominates the opacityand is also the raw material at the base of any chain of agglomeration into planetismals, only represents about 1 %of the mass of the disk. The primary gas component is more difficult to observe, although there has been increasingsuccess with the exploitation of various atomic and molecular lines. The general picture for the disk geometry, at leastin its initial stage, gives a flared structure in stable Keplerian rotation with an exponential density taper that producesa soft outer edge at a radius anywhere between 30–200 AU8 from the star.

However it is critical to note, particularly in the context of planetary assembly processes, that protoplanetary disksare very dynamic places. Cluster studies imply that the median time for dissipation of the disk is around 3 Myr withvery few surviving as long as 10 Myr, which serves to place strict limits on the time available for the formation ofplanets (for binary stars, higher mass stars, or low metallicity environments, timescales are dramatically shorter). Fur-thermore, both the global structure and the chemistry of the disk evolve dramatically, driven by a host of processessuch as viscous accretion and photo-evaporation. Of central relevance is the fate of the dust, which undergoes signifi-cant evolution as icy mantles grow from the gas phase and grains stick together reaching mm or cm size scales9, whichin turn causes settling to the midplane and so promotes higher densities and further growth.

The net effect of all these processes is a decrease in the accretion rate which eventually falls below the photo-evaporation rate so that the inner disk drains leaving a cavity between the star and outer disk, terminating furtheraccretion. Several classes of object exhibiting a bewildering diversity of observational characteristics populate the no-mans-land between the era of active star formation and a mature solar system. Among the most intenstively studiedare the so-called transition disks whose defining characteristic is a lack of near-IR emission but with a strong excessat longer wavelengths (> 10 µm). Significantly complicating the evolutionary picture, it has been found that a subsetof transition disks do maintain some accretion10, an anomaly which requires further physical processes to explain.With millimeter telescopes now delivering beautifully resolved images of disks with large inner cavities, there aretwo leading explanations for how such gaps may be carved in a disk: grain growth or dynamical interactions with amassive body. Because these two should generate distinct observational signatures, there is mounting evidence for asignificant population of objects for which the favored explanation requires planetary mass bodies within the disk10.

Figure 1: Left Panel: Recent 3D hydrodynamical modeling finds that under some circumstances protoplanetary diskscan be unstable to gravitational collapse with the resulting fragmentation generating gravitationally bound clumps withmasses corresponding to those of gas giant planets11. The simulation also depicts many phenomena associated withastrophysical disks such as gravitational wakes, spiral density waves and the formation of disk gaps. Right Panel:Near-infrared (H-band) polarised intensity image of the disk surrounding the young stellar system AB Aurigae takenwith the HiCIAO camera at the Subaru Telescope. The central hole is due to the coronagraph masking the brightcentral star, and a differential dual-beam polarimetry technique12 was also employed. A wealth of structure has beenextracted from such images including two major rings, at least 8 spiral arms, gaps, clumps, knots and evidence fordisk warping.

Hydrodynamical modelling of gravitational collapse of protoplanetary disk, with resulting fragmentation

causing gravitationally bound clumps Mayer L., et al., 2007, ApJ 661 77

SCIENCE CASE: IMAGING THE INNER MASS-LOSS REGION OF EVOLVED STARS

Mass-loss process poorly understood • Ejected material comprises next generation of

stars & planets • What drives this wind? • Directly image inner dust condensation region,

explore grain size and composition (polarimetry)

Origin of asymmetries • Asymmetries observed at wide scales - where do

they originate? • At star (effect of stellar atmosphere) or interstellar

medium?

Extension of red giant atmospheres • Does radiative pressure on dust in atmosphere

contribute to extension? Peter Woitke: http://www.strw.leidenuniv.nl/~woitke/AGB_popular.html

W Hya light curve (AAVSO)

New mode - Differential spectral line imaging - Hɑ, SII Operates double-differential to calibrate NCP

Diffraction limited imaging through atmosphere: One method: Non-Redundant Masking

State of the art 10m telescope: - Diffraction limit ~ 10 mas - Seeing limited ~ 1000 mas The goal: mitigate the effect of the turbulent atmosphere to allow diffraction limited, high contrast imaging.

Aperture masking interferometry present state of art —> Each and every hole-pair is baseline of interferometer —> Recover diffraction limited performance

ψ - Measured baseline phaseφ - Actual baseline phasee - error from atmosphere

ψ12 = φ12 + e2

ψ23 = φ23 + e3 - e2

ψ31 = φ31 - e3

C.P. = ψ12 + ψ23 + ψ31 = φ12 + φ23 + φ31

12

3

Visibilities - a power spectrum, independent of phase

Closure phase independent of phase

WHY AO & ADVANTAGE OF HIGH TIME-CADENCE

Poor seeing / large residual WF errorHigh time cadence = high fringe visibilities (and S/N)

DIFFERENTIAL CALIBRATION - IN AN IDEAL WORLD…

H polarisation image⭑

Error (seeing, systematics)

H polarisation visibilitiesx

Error visibilities

FT

V polarisation image⭑

Error (seeing, systematics)

V polarisation visibilitiesx

Error visibilities

FT

÷ = V/H polarisation visibilities (CLEAN OBSERVABLE)

DIFFERENTIAL CALIBRATION - TRIPLE LAYER SWITCHING!

Half-wave plate

Differential Observable

VH / VV&

CPH - CPV

H / V

H / V

V / H

Prism

H

V

LCVR

Stat

e 1

0°

45°

H / V

V / H

Calibrate

Calibrate

Calibrate

Calibrate

Calibrate

Calibrate

Calibrate

H / V

V / H

H = horizontally polarised visibilitiesV = vertically polarised visibilitiesHalf-wave plate

Stat

e 2

PrismH

V

Prism

LCVR

Stat

e 1

Stat

e 2

Prism

H

V

H

V

Address non-common-path & non-simultaneous errors

DIFFERENTIAL CALIBRATION - TRIPLE LAYER SWITCHING!

Address non-common-path & non-simultaneous errors

To allTo EMCCD

To EMCCD

MaskWheel

QW

P

SCExAO optics

To PyWFSPeriscope M2

Periscope M1

BRT L2

Pupil Plane

FilterWheel

TCLCC

Computer

FLC EMCCDCamera

BRT L1

Cam L1 Cam L2Dicr. M2

Dicr. M1 OAP2 OAP1DM

Translation stageRotation stage

To coronagraphs,IR cameras

Telescope(M1, M2, M3)

IP Cal. Source

Supe

r-K

OAP

col

.

Dep

olz

ArduinoSMF

Lin

polz

AO 188QW

P

Polarising /SpectralBeamsplitter

HW

P

Lin

Pol

EMCCDCamera

Differential filter-wheel

Instrumental Polarisation Calibration + Compensation• Measurement of instrumental polarisation is important especially due to k-

mirror (image rotator) in AO188, periscope within SCExAO• Calibration sets using AO188 cal lamp + SCExAO + VAMPIRES have been measured, to

constrain instrumental Mueller matrix (then multiply model by M-1)

• But don’t measure V - lose signal, can’t correct by calibration• Use Eigenpolarisation approach

Corrects for most of instrumental polarisation using two quarter-wave plates

On the right is a measured chi-squared map showing the accuracy of Stokes Q transmission through AO188+SCExAO+VAMPIRES as a function of QWP angles. The minimum (blue) corresponds to the setting with optimally corrected transmission.

Final calibration should be confirmed each observing period by polarised standard star observations.

NECESSARY TECHNICAL NOTE…Effective way of plotting calibrated VAMPIRES data is with a polarised visibility ratio plot:

a)

b) c)

0 7.22.4 4.8Baseline Length (m)

Synthetic example data

Y Axis: Ratio of horizontally-

polarised to vertically-polarised interferometric visibilities (Fourier power)

X Axis: Azimuthal angle (on-sky) of the baseline (spatial frequency coordinate)

Color scale: Baseline length (on-sky) (abs value

value of spatial frequency coordinate)

2 plots:Correspond to Stokes Q

and U coordinates

Should = 1 for all unpolarised targets…Demonstrates calibration precision

NRM MODE CALIBRATION PERFORMANCE➤ Demonstrated polarised-differential NRM visibility calibration

to 1 part in 10^3 per baseline. ➤ Roughly speaking, achievable contrast at 0.5𝜆/D (10 mas) goes as √nBaselines - DOF

Conventional 2-beam polarimetry Triple-differential calibration

NRM-MODE EXAMPLE SCIENCE-RESULTInvestigation of supergiant mass-loss via direct imaging of Circumstellar dust around Red Supergiant μ Cephei

• Observed with annulus mask at 775 nm• Raw differential polarized visibilities show

distinctive sinusoidal signature of circumstellar dust shell, with clear asymmetry (right)

• Dust scattering model fitted via synthetic differential visibilities (below)

NRM-MODE EXAMPLE SCIENCE-RESULTModel-fitting reveals extended, asymmetric dust shell, originating within the outer stellar atmosphere, without a visible cavity. Such low-altitude dust (likely Al2O3) important for unexplained extension of RSG atmospheres.

Left: model image, shown in polarized intensity. Middle: model image show in four polarisations. Right: Model image (intensity), shown with wide field MIR image (from de Wit et al. 2008 – green box shows relative scales. Axis of extension in MIR image aligns with the close-in VAMPIRES image.

Inner radius: 9.3 ± 0.2 mas (which is roughly Rstar) Scattered-light fraction: 0.081 ± 0.002

PA of major axis: 28 ± 3.7 ° • Aspect ratio: 1.24 ± 0.03

IMAGING/PDI MODE➤ High-speed acquisition and polarisation switching rate (up to 500

Hz) - allows offline tip/tilt correction and ‘lucky imaging’ approach, unlike CCD-shuffle based systems

➤ Example: 750 nm PSF, 100 frames/sec, discard 80% lowest-Strehl frames, align and sum (~30 secs of data)

IMAGING-MODE EXAMPLE SCIENCE-RESULT➤ Direct imaging of omi Cet - clear asymmetry seen.

➤ Speckle imaging - diffraction-limited

6 mas / pxFWHM = ~18 mas𝛌 / D = 19.3 mas

omi Cet

Observed 12/09/2017

PSF ref (63 Cet)

100 mas 100 mas

IMAGING-MODE EXAMPLE SCIENCE-RESULT

6 mas / pxFWHM = ~18 mas𝛌 / D = 19.3 mas

Observed 12/09/2017

PSF ref (63 Cet)

omi Cet

100 mas

IMAGING-MODE EXAMPLE SCIENCE-RESULT➤ Observation of omi Cet - dust shells and disk around omi Cet b observed

➤ Very preliminary reduction

Stokes Q Stokes UObserved 12/09/2017

IMAGING-MODE EXAMPLE SCIENCE-RESULT➤ Observation of omi Cet - dust shells and disk around omi Cet b observed

➤ Very preliminary reduction

Stokes Q Stokes U

omi Cet b disk omi Cet b disk

Inner shell ~ 2R★ Gap & 2nd shell Inner shell ~ 2R★

Gap & 2nd shell

Dust transfer?

Observed 12/09/2017

IMAGING-MODE EXAMPLE SCIENCE-RESULT➤ Effect of single vs triple polarisation calibration

Stokes Q

3-layer PDI

IMAGING-MODE EXAMPLE SCIENCE-RESULT

Stokes Q Stokes U

Disk aroundAB Aur

100 mas

ConclusionVAMPIRES uses aperture masking & differential polarimetry:Directly image dusty inner circumstellar regions (protoplanetary disks, evolved star mass-loss) in scattered light • Protoplanetary disks, evolved star mass-loss• Resolution ~10 mas, 600 - 800 nm

Incorporated into SCExAO extreme AO system at the Subaru 8m telescope • Simultaneously conducts R,I band observations at same time as IR instruments (CHARIS, …)• Extreme AO correction, access to telemetry• Differential line imaging - Hɑ, OI, SII• Can operate double-differential to calibrate some NCP• Non-redundant masking and full pupil (inc. speckle, PDI) modes

On-sky results • Differential visibilities to 0.1% and closure phases to ~0.5 deg• Evolved star and protoplanetary disk science results in prep for publication

Recent upgrades • Fast FLC polarisation switching and dual cameras (up to ~500 fps)• Spectral differential as well as polarisation modes• Time-resolved differential polarimetry (suitable for PDI-speckle and NRM)

Available for use! • Apply via usual Subaru proposals… or talk to me!

VAMPIRES !Visible Aperture Masking Polarimetric Interferometer for Resolving Exoplanetary Signatures

What is VAMPIRES?!VAMPIRES is a new instrument which will image the innermost region of protoplanetary disks, with scattered light, at unprecedented resolution (~ 10 mas). Combining the power of aperture masking interferometry with differential polarimetry, VAMPIRES perfectly complements coronographic observations. Integrated into the SCExAO extreme-AO system at the Subaru telescope, it can conduct observations at visible wavelengths simultaneously with near-IR HICIAO observations.

Unlocking the mystery of planet formation!How the disks of gas and dust surrounding young stars evolve into planetary systems - like our own solar system - is an outstanding question today. The inner 10’s of AU of protoplanetary disks is key to understanding planet formation and disk structure, but is difficult to resolve with standard direct imaging and coronagraphic techniques. VAMPIRES will allow • Precise measurement of transition disk gaps and inner disk -

Including density maps of dust within gap • Detection and imaging of density perturbations from massive bodies within

disk (e.g. warps, spirals) • Imaging and measurement of gravitationally bound clumping dust

immediately surrounding forming planets • Spatially resolved measurement of dust grain size and species

An innovative technique!VAMPIRES has high resolutions (~10 max) but small field-of-view (~500 mas), perfectly complementing coronagraphic observations. VAMPIRES combines two techniques:

BackgroundCircumstellar disks naturally arise in simulations of the gravitational collapse of large clouds due to the conserva-

tion of angular momentum. There are many lines of independent evidence (such as the regularity of our solar systemnoted by Kant) which point to such disks being the cradle for planetary birth. In recent years a picture for some, butby no means all, of the steps needed for the formation of a mature solar system has been emerging.

Within a typical timescale of 0.5 Myr6 a visible star emerges from the dusty cocoon of its birth (evolving from aClass I to Class II young stellar object), and accretion onto the central star is effectively complete leaving a remnantdisk with only about⇠5 MJupiter (or about 1 % of the mass of the central star)7. The dust, which dominates the opacityand is also the raw material at the base of any chain of agglomeration into planetismals, only represents about 1 %of the mass of the disk. The primary gas component is more difficult to observe, although there has been increasingsuccess with the exploitation of various atomic and molecular lines. The general picture for the disk geometry, at leastin its initial stage, gives a flared structure in stable Keplerian rotation with an exponential density taper that producesa soft outer edge at a radius anywhere between 30–200 AU8 from the star.

However it is critical to note, particularly in the context of planetary assembly processes, that protoplanetary disksare very dynamic places. Cluster studies imply that the median time for dissipation of the disk is around 3 Myr withvery few surviving as long as 10 Myr, which serves to place strict limits on the time available for the formation ofplanets (for binary stars, higher mass stars, or low metallicity environments, timescales are dramatically shorter). Fur-thermore, both the global structure and the chemistry of the disk evolve dramatically, driven by a host of processessuch as viscous accretion and photo-evaporation. Of central relevance is the fate of the dust, which undergoes signifi-cant evolution as icy mantles grow from the gas phase and grains stick together reaching mm or cm size scales9, whichin turn causes settling to the midplane and so promotes higher densities and further growth.

The net effect of all these processes is a decrease in the accretion rate which eventually falls below the photo-evaporation rate so that the inner disk drains leaving a cavity between the star and outer disk, terminating furtheraccretion. Several classes of object exhibiting a bewildering diversity of observational characteristics populate the no-mans-land between the era of active star formation and a mature solar system. Among the most intenstively studiedare the so-called transition disks whose defining characteristic is a lack of near-IR emission but with a strong excessat longer wavelengths (> 10 µm). Significantly complicating the evolutionary picture, it has been found that a subsetof transition disks do maintain some accretion10, an anomaly which requires further physical processes to explain.With millimeter telescopes now delivering beautifully resolved images of disks with large inner cavities, there aretwo leading explanations for how such gaps may be carved in a disk: grain growth or dynamical interactions with amassive body. Because these two should generate distinct observational signatures, there is mounting evidence for asignificant population of objects for which the favored explanation requires planetary mass bodies within the disk10.

Figure 1: Left Panel: Recent 3D hydrodynamical modeling finds that under some circumstances protoplanetary diskscan be unstable to gravitational collapse with the resulting fragmentation generating gravitationally bound clumps withmasses corresponding to those of gas giant planets11. The simulation also depicts many phenomena associated withastrophysical disks such as gravitational wakes, spiral density waves and the formation of disk gaps. Right Panel:Near-infrared (H-band) polarised intensity image of the disk surrounding the young stellar system AB Aurigae takenwith the HiCIAO camera at the Subaru Telescope. The central hole is due to the coronagraph masking the brightcentral star, and a differential dual-beam polarimetry technique12 was also employed. A wealth of structure has beenextracted from such images including two major rings, at least 8 spiral arms, gaps, clumps, knots and evidence fordisk warping.

Hydrodynamical modelling of gravitational collapse of protoplanetary disk, with resulting fragmentation

causing gravitationally bound clumps - Mayer L., et al., 2007, ApJ 661 77

1. Aperture masking for super-diffraction-limited resolution through the turbulent atmosphere!Aperture masking converts Subaru’s single 8m pupil into an array of smaller sub-pupils, forming an interferometer. Analysis of the interference patterns on the detector allows two key observables - visibilities (power spectrum) and closure phase - to be recovered. These observables are unaffected by seeing, and enable the full diffraction limited resolution of the 8m telescope.

Visibilities!The power spectrum

Interferogram

Power spectrum

Closure phases!Vector sum of phases around closed triangle - unaffected

by atmospheric phase errors

Fo

urie

r Transfo

rm

ψ - Measured baseline phase"φ - Actual baseline phase"e - error from atmosphere "

ψ12 = φ12 + e2"ψ23 = φ23 + e3 - e2"

ψ31 = φ31 - e3"

C.P. = ψ12 + ψ23 + ψ31 = φ12 + φ23 + φ31

2

3

1

VAMPIRES field-of-view and resolution perfectly complements coronagraphy.

MWC758 - Base image: Grady et al., ApJ, 2013

2. Differential polarimetry for super-precise calibration and polarimetric measurements!Precise calibration is vitally important in interferometry, in order to remove systematic errors and obtain high precision observables. This in turn allows detection of high contrast objects at high spatial resolution. In VAMPIRES, calibration is performed between orthogonal polarisations, in nested layers. By switching polarisation channels with a liquid crystal variable retarder and a half-wave-plate, systematics arising from the instrument are cancelled out, and the astrophysical signal remains.

Wollaston Prism

LCVRFromTelescope (M1, M2)

To SCExAO coronagraph and HiCIAO

Camera

Longpass dichroic mirror

< 1 m

Mask Wheel

Half-wave plate

> 1

m

Dichroics

Pyramid WFS

SCExAO 2KDeformable Mirror

AO188

Computer

EMCCD Camera

Timing Micro-

controller

Filter Wheel

Various Dedicated

Controllers

Calibration Source

Super-K

De-polz

Lin Polz

WMKRM½IW�SRI�SV�QSre optical elements

(lenses or mirrors)

Pupil plane

QW

P

M3

Half-wave plate

Differential Observable

VH / VV&

CPH - CPV

H / V

H / V

V / H

Wollaston

Wollaston

H

V

LCVR

VoltHI

VoltLO

Wollaston

H

V

Wollaston

H

V

LCVR

VoltHI

VoltLO

0°

45°H

V

H / V

V / H

Calibrate

Calibrate

Calibrate

Calibrate

Calibrate

Calibrate

Calibrate

H / V

V / H

e.g.

[(H/V) / (V/H)]0.5 = (H/V)

Triple differential calibration processH = horizontally polarised visibilitiesV = vertically polarised visibilitiesH / V = polarised visibility ratio[Analogous process with closure phases]

Half-wave plate

On-sky calibration results!VAMPIRES observations of standard stars in July 2013 demonstrated VAMPIRES calibration precision. These observations were made with short (~1 minute) total integrations and with conventional AO (AO188). Accordingly, we expect these results to improve even further with longer integrations and the commissioning of the SCExAO closed-loop extreme-AO operation. The goal is to have visibility RMS of ~0.1% and closure phase RMS of <<1 degree.

Simulated data!A hypothetical transition disk was modelled using MCFOST (Pinte, et al., 2006, A&A 459). The disk was based on the geometric parameters of IRS 48, placed at a distance such that Rcav ~ 30 max, and the mean surface polarisation was 20%. The model was calculated at a wavelength of 700 nm.

Baseline azimuth angle (radians)1 2!0!-1-2Po

laris

ed v

isib

ility

ratio

1.0

1.1

0.9

σ = 6.8 %

1 Layer: Wollaston Prism Small temporal errors (small error bars), large

systematics (from non-common-path)

Baseline azimuth angle (radians)1 2!0!-1-2Po

laris

ed v

isib

ility

ratio

1.0

1.1

0.9

σ = 2.4 %

1 Layer: Liquid Crystal Larger temporal errors, low systematics (no non-

common-path)

Baseline azimuth angle (radians)1 2!0!-1-2Po

laris

ed v

isib

ility

ratio

1.0

1.1

0.9

3 Layers: Wollaston Prism + LCVR + HWPSystematic errors from spatially dependent

instrumental polarisation largely suppressedσ = 0.42 %

NB Still limited by random error (photon, EM & read noise), and no Extreme-AO

0 Layers: Raw closure phase

1 Layer: Liquid Crystal

2 Layers: Liquid Crystal + HWP

Barnaby Norris1, Peter Tuthill1, Guillaume Schworer1, Paul Stewart1, Nemanja Jovanovic2, Olivier Guyon2, Frantz Martinache2

1. Sydney Institute for Astronomy, University of Sydney, A28 Physics Rd. University of Sydney NSW 2006, Australia. 2. Subaru Telescope, National Astronomical Observatory of Japan, 650 North A'ohoku Place, Hilo, HI 96720, U.S.A.

Intensity map of whole disk, with inner region 'observed' shown.

Intensity map of inner region Differential visibility signal of the inner disk region as seen by VAMPIRES, for Stokes Q (left) and Stokes U (right)

Inner region seen in 4 linear polarisations